catchment influence on nitrate and dissolved organic

Catchment influence on nitrate and dissolvedorganic matter in Alaskan streams acrossa latitudinal gradientTamara K. Harms1,2, Jennifer W. Edmonds3, Hélène Genet1, Irena F. Creed4, David Aldred4,Andrew Balser5, and Jeremy B. Jones1,2

1Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, Alaska, USA, 2Department of Biology and Wildlife,University of Alaska Fairbanks, Fairbanks, Alaska, USA, 3Department of Physical and Life Sciences, Nevada State College,Henderson, Nevada, USA, 4Department of Biology, Western University, London, Ontario, Canada, 5Climate Change ScienceInstitute, Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA

Abstract Spatial patterns in carbon (C) and nitrogen (N) cycles of high-latitude catchments have beenlinked to climate and permafrost and used to infer potential changes in biogeochemical cycles underclimate warming. However, inconsistent spatial patterns across regions indicate that factors in addition topermafrost and regional climate may shape responses of C and N cycles to climate change. We hypothesizedthat physical attributes of catchments modify responses of C and N cycles to climate and permafrost. Wemeasured dissolved organic C (DOC) and nitrate (NO3

�) concentrations, and composition of dissolvedorganic matter (DOM) in 21 streams spanning boreal to arctic Alaska, and assessed permafrost, topography,and attributes of soils and vegetation as predictors of stream chemistry. Multiple regression analysesindicated that catchment slope is a primary driver, with lower DOC and higher NO3

� concentration in streamsdraining steeper catchments, respectively. Depth of the active layer explained additional variation inconcentration of DOC and NO3

�. Vegetation type explained regional variation in concentration andcomposition of DOM, which was characterized by optical methods. Composition of DOM was furthercorrelated with attributes of soils, including moisture, temperature, and thickness of the organic layer.Regional patterns of DOC and NO3

� concentrations in boreal to arctic Alaska were driven primarily bycatchment topography and modified by permafrost, whereas composition of DOM was driven by attributesof soils and vegetation, suggesting that predicting changes to C and N cycling from permafrost-influencedregions should consider catchment setting in addition to dynamics of climate and permafrost.

1. Introduction

High-latitude catchments play a significant role in global elemental cycling by storing significant quantities oforganic carbon (C) and nitrogen (N) in cold, wet soils [Harden et al., 2012; Tarnocai et al., 2009]. Large fluxes ofdissolved materials link high-latitude catchments to coasts [Holmes et al., 2012; McGuire et al., 2009], wheredissolved organic C (DOC) and N may fuel microbial respiration or stimulate productivity in the ArcticOcean [Carmack et al., 2004; Holmes et al., 2008; Le Fouest et al., 2013]. Furthermore, significant gaseous fluxesof C result from respiration of DOC in aquatic ecosystems of high-latitude catchments, whichmay constitute afeedback to further climate warming [Kling et al., 1991; Striegl et al., 2012]. Recent observations of high-latitude catchments indicate decreased flux of DOC and increased flux of nitrate (NO3

�) from an arctic riverat a decadal time scale [McClelland et al., 2007]. Export of DOC has also decreased relative to discharge overseveral decades in the Yukon River Basin, which drains predominantly discontinuous permafrost [Dornblaserand Striegl, 2007; Striegl et al., 2005]. Additionally, upland headwater catchments in the same basin yieldnet export of N as NO3

� [Jones et al., 2005]. Despite these coincident trends, evidence for changing C and Ncycles is not consistent across all high-latitude catchments. For example, whereas a latitudinal decline in streamNO3

� was observed across a gradient spanning discontinuous to continuous permafrost in Alaska [Jones et al.,2005], no latitudinal pattern in inorganic N was observed across a similar latitudinal gradient in Siberia [Freyet al., 2007]. Similarly, hypotheses and observations from studies of diverse regions and spatial extents disagreeas to whether DOC fluxes will increase or decrease in response to thawing permafrost [Frey and Smith, 2005;Laudon et al., 2012; McClelland et al., 2007; Striegl et al., 2005].

Permafrost has a dominant influence on hydrologic and biogeochemical cycles at high latitudes. Frozensoils restrict flow of water to seasonally thawed surface horizons, termed the active layer, and limit

HARMS ET AL. NITRATE AND DOM IN HIGH-LATITUDE STREAMS 350

PUBLICATIONSJournal of Geophysical Research: Biogeosciences

RESEARCH ARTICLE10.1002/2015JG003201

Key Points:• Deeper active layer correlated withless dissolved organic carbon instreams draining permafrost

• Higher nitrate and lower organiccarbon concentrations in streamsdraining steep catchments

• Vegetation and soils influencecomposition of dissolved organicmatter in Alaskan streams

Supporting Information:• Supporting Information S1

Correspondence to:T. K. Harms,[email protected]

Citation:Harms, T. K., J. W. Edmonds, H. Genet,I. F. Creed, D. Aldred, A. Balser, andJ. B. Jones (2016), Catchment influenceon nitrate and dissolved organic matterin Alaskan streams across a latitudinalgradient, J. Geophys. Res. Biogeosci., 121,350–369, doi:10.1002/2015JG003201.

Received 27 AUG 2015Accepted 6 JAN 2016Accepted article online 10 JAN 2016Published online 5 FEB 2016Corrected 25 FEB 2016

This article was corrected on 25 FEB2016. See the end of the full text fordetails.

©2016. American Geophysical Union.All Rights Reserved.

http://publications.agu.org/journals/

http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2169-8961

http://dx.doi.org/10.1002/2015JG003201

http://dx.doi.org/10.1002/2015JG003201

http://dx.doi.org/10.1002/2015JG003201

http://dx.doi.org/10.1002/2015JG003201

http://dx.doi.org/10.1002/2015JG003201

groundwater-surface water interactions [Carey and Quinton, 2005]. This results in high concentration ofDOC in receiving waters of catchments underlain by permafrost relative to those with little or no perma-frost [Balcarczyk et al., 2009; Neff et al., 2006; Petrone et al., 2006]. Accordingly, composition of DOM inhigh-latitude streams typically reflects vegetation sources during snowmelt, when flowpaths are shallow,with increased evidence of decomposition at base flow [O’Donnell et al., 2010; Spencer et al., 2008;Walker et al., 2013]. Dissolution and transport of inorganic solutes, including N, result when permafrostthaw yields a thicker active layer, permitting deeper flowpaths through catchments [Harms and Jones,2012; Keller et al., 2010; Keuper et al., 2012; Lafrenière and Lamoureux, 2013; Walvoord and Striegl, 2007].Thus, thawing at the southern extent of permafrost may contribute to an observed latitudinal decline inNO3

� concentration from subarctic to arctic streams [Jones et al., 2005]. Despite these general spatialand seasonal patterns linking permafrost-driven hydrology to mobilization of C and N from catchments,significant variation in concentration or load and composition of DOM and inorganic N occurs across catch-ments and among regions at high latitudes [Holmes et al., 2012; McClelland et al., 2014; Roth et al., 2013;Tank et al., 2012; Walker et al., 2013].

In addition to climate-driven permafrost dynamics, temperature influences catchment biogeochemistrythrough its effects on rates of geochemical weathering and biological processes, and the composition orcover of vegetation. Across a broad suite of Siberian catchments encompassing both permafrost andpermafrost-free regions, concentration of DOC in streams increased as a function of mean annual airtemperature, which suggests potential for significant increases in delivery of DOC to coasts under continuedclimate warming [Frey et al., 2007]. Temperature also has a dominant effect on the composition of DOM,through its influence on rates of decomposition [Williams et al., 2010]. Nitrate concentration in forested bor-eal catchments correlates positively with mean summer or annual air temperature [Kortelainen et al., 2006;Sponseller et al., 2014], which is likely due to climate effects on the distribution and productivity of vegetation.Additionally, species-specific acquisition of nutrients, nutrient use efficiencies, and productivity of plants canalter abundance of C and N available for export, thereby influencing solute concentrations in receiving waters[Judd and Kling, 2002; Lovett et al., 2002].

Independent of climatological setting, catchment topography explains variation in concentration and exportof dissolved C and N in streams and lakes, due to its strong influence on water residence time [McGuire et al.,2005]. Longer water residence times, correlated with low relief and presence of wetlands, typically corre-spond with increased concentrations of organic solutes and decreased concentration of inorganic solutes,particularly NO3

� [Clow and Sueker, 2000; Creed et al., 2008; Dillon and Molot, 1997; Gergel et al., 1999;Sponseller et al., 2014; Watmough et al., 2004]. Low relief may also yield characteristic organic molecules,due to effects of soil moisture on rates of decomposition or distinct plant communities within wetlands[Kothawala et al., 2015]. For example, greater relative protein-like and humic components of DOM, detectedby analysis of fluorescence attributes of streamwater, have been observed in wetland-influenced catchmentscompared to upland catchments [Strohmeier et al., 2013; Williams et al., 2010]. Protein-like components arepositively correlated with decomposition of DOC [Balcarczyk et al., 2009; Fellman et al., 2009], and thus, catch-ment characteristics in turn may influence the eventual fate of exported DOM.

Catchment characteristics influence the character and eventual fate of exported C and N, but the specificattributes of high-latitude catchments that determine export of C and N remain unclear [McClelland et al.,2015]. Predicting response of high-latitude catchments to climate is critical because dissolved fluxes maycontribute substantial, but less frequently evaluated components of terrestrial C and nutrient budgets, andlarge fluxes of C and nutrients link high-latitude catchments to coasts. Due to limited data available for para-meterization of models and gaps in underlying conceptual frameworks, existing biogeochemical models spe-cific to high-latitude regions do not accurately represent transfer of solutes from terrestrial to aquaticecosystems, yielding estimates of solute exports to coasts that vary widely among models [Holmes et al.,2012; Kicklighter et al., 2013]. Increased understanding of the factors driving spatial and temporal heteroge-neity in dissolved C and N exports will enable refined predictions of material exports and modeled responsesto changing climate. Therefore, the objectives of this study were to identify the characteristics of catchmentsthat influence concentrations of NO3

� and DOC, and composition of DOM in streams across a latitudinal gra-dient in Alaska spanning discontinuous to continuous spatial extent of permafrost. Specifically, we assessedrelationships between stream chemistry and catchment characteristics including topography, vegetation,permafrost, and climate. We hypothesized that permafrost extent and active layer depth establish regional

Journal of Geophysical Research: Biogeosciences 10.1002/2015JG003201


patterns in stream chemistry due to strong influence on catchment hydrology and vegetation. Alternatively,we hypothesized that as in permafrost-free regions, catchment topography is correlated with streamchemistry, due to the relationship between catchment slope and water residence time.

2. Methods2.1. Site Description

We sampled streams across a north-south gradient spanning approximately 4° latitude (Figure 1, Table 1, andTable S1 in the supporting information). Samples were collected three times (May, July, and August) in 2009,corresponding to snowmelt, summer, and early autumn periods, respectively. In May, sampling occurredprior to peak discharge from the northernmost streams and in smaller catchments throughout the gradient,whereas sampling coincided with snowmelt recession in the largest of the southernmost streams (Figure S1).Except for the Chena, Chatanika, and Salcha Rivers, ice was present in all channels during sampling in May. Allstreams were sampled at or near base flow conditions in July and August (Figure S1). Additional samples were

Figure 1. Study streams arrayed along a latitudinal gradient. Shading represents estimated permafrost extent.



collected from a subset of streams in 2010 for verification of the parallel factor model of DOM composition(see excitation-emission matrix (EEM) methods below). The gradient encompassed streams underlain by dis-continuous permafrost in the south, transitioning to continuous permafrost in the north. Although glacialstreams occurred along the latitudinal transect, particularly within the Brooks Range, these streams are lessstrongly coupled to the surrounding watershed, and thus were not included in analyses. All catchmentsare within the Yukon River Basin, except for three streams on the north slope of the Brooks Range, whichrepresent tundra ecosystems underlain by continuous permafrost. Anthropogenic development accountedfor a maximum of 0.7% of catchment area across the catchments included in the study. Fertilization experi-ments near the northern and southern ends of the latitude gradient have indicated that terrestrial primaryproduction is typically limited by N [Chapin et al., 1995; Yarie and Van Cleve, 2010].

2.2. Stream Water Sampling and Analysis

We collected triplicate water samples from each stream and filtered samples (0.7μm glass fiber filters) at thetime of collection. Samples were then frozen until analysis. We analyzed NO3

� on a Dionex IC25 ion chroma-tograph (limit of quantitation: 0.2μM) and DOC by nondispersive infrared detection on a Shimadzu TOC-5000(limit of quantitation: 8μM).

2.3. Composition of Dissolved Organic Matter

We measured absorbance of UV and visible light, and fluorescence spectra of stream water to quantifycomposition of DOM. Previous research has demonstrated that DOM from varying sources or ages producescharacteristic fluorescence patterns, which can be used to infer patterns in DOM complexity and composition[Coble, 2007]. Excitation-emission matrices (EEMs) were generated from fluorescence spectra by excitingsamples across a range of wavelengths of light and monitoring intensity and wavelengths of resulting lightemissions. Characteristic excitation and emission wavelengths vary due to the composition of organicmolecules present in the sample. EEMs were collected for replicate surface water samples from all sites inMay, July, and August in 2009, and a subset of streams were sampled again in 2010 to increase power ofthe Parallel Factor Analysis (PARAFAC) model (see below). Fluorescence spectra were measured on aPerkin-Elmer LS 55 fluorescence spectrometer. Excitation wavelengths ranged from 250 to 400 nm in10 nm increments, and the corresponding emission spectrum was collected from 350 to 500 nm at2 nm increments.

Table 1. Catchment Attributes Used as Predictors in Models of Stream Chemistrya

Attribute Mean Minimum Maximum

Catchment physical setting (n = 20)Slope (deg) 8.97 3.02 14.92Wetlands (% of catchment area) 8.47 2.00 20.51Aspect (deg from 0) 88.87 74.41 108.56Stream length:catchment length (�) 11.9 0.7 34.7Catchment area (km2) 858.9 3.9 5726.4Vegetation cover (proportional) (n = 18)Wetland species 0.07 0.02 0.25Upland deciduousb 0.09 0.00 0.37Upland blank spruce 0.35 0.11 0.53Upland tundra 0.13 0.00 0.53TEM-modeled soil attributesc (n = 18)Active layer depth (m) 1.33 1.02 1.61Organic layer thickness (m) 0.14 0.10 0.18Volumetric water content, May (m3/m3) 0.63 0.41 0.81Volumetric water content, July (m3/m3) 0.42 0.26 0.50Volumetric water content, August (m3/m3) 0.57 0.33 0.68Soil temperature, May (°C) 4.8 2.4 8.1Soil temperature, July (°C) 11.6 7.8 16.2Soil temperature, August (°C) 8.5 5.8 12.2

aN represents the number of catchments for which data were available.bLowland vegetation cover classes were additionally included in candidate models.cVolumetric water content and temperature are summarized for the shallow organic layer.



EEMs were analyzed by Parallel Factor Analysis (PARAFAC) with the N-way toolbox for MATLAB [Anderssonand Bro, 2000]. Standard methods were used to correct emission intensities for instrument bias, daily varia-tion in instrument performance (normalization using the water-Raman curve), and the inner-filter effect[Cory et al., 2010]. Fluorescence of a Milli-Q water blank was subtracted from each sample. PARAFAC analysisconsiders the entire sample set of EEMs and identifies repeated patterns in high fluorescence (“hot spots”)within the three-dimensional data matrix. These hot spots are labeled components, which can then beassigned in a unique proportion to each sample. Components can be compared with previously reportedmodel results to identify the general class of organic molecules that created the characteristic fluorophoreswithin the component’s region of the matrix (Table 3). Model validation was based on split-half analysis usingthe DOMFluor toolbox [Stedmon and Bro, 2008], in which each sample is randomly assigned to one of twoseparate data sets, followed by independent evaluation of a prescribed number of PARAFAC components.The results from these analyses are compared, and if deviation in component characteristics is below a setthreshold for the two data sets, the model is considered validated.

In addition to components resulting from PARAFACmodel analysis, we calculated the fluorescence index (FI),which typically correlates with the source of DOM. Specifically, higher values of FI indicate microbial origin ofmolecules contributing to the DOM pool, whereas lower values indicate a largely terrestrial origin from higherplants [Cory et al., 2010; Williams et al., 2010]. FI is most commonly calculated as the ratio of emittedfluorescence intensity at 470nm to the intensity at 520nm with an excitation wavelength of 370nm (f470/f520)[Jaffe et al., 2008]. However, due to the instrumentation used in this study, we calculated FI using data with anemission intensity ratio of 450nm/500nm, with an excitation wavelength of 370nm [McKnight et al., 2001].Although this estimate of FI limits direct comparison of values with other studies, the relative magnitude ofthe FI values is consistent and can be interpreted in the same manner as for FI calculated from longer wave-lengths [Cory et al., 2010].

Absorbance of UV and visible light was measured on a Shimadzu UV-1700 UV-Visible spectrophotometer, atwavelengths from 250 to 500nm in 1nm increments. Specific UV Absorbance at 254nm (SUVA254) was calcu-lated by dividing absorbance at 254nm by the DOC concentration (mg C/L) and is a measure of aromaticity[Weishaar et al., 2003]. The slope ratio (SR) was quantified as the ratio of the log-transformed spectral slopes ofabsorbance at wavelengths 275–295nm and wavelengths of 350–400nm. SR is related to molecular weight ofcompounds in the DOM pool, with higher values corresponding to lower molecular weights [Helms et al., 2008].

2.4. Spatial Data Collection and Analysis

Topographic data were extracted from a 60m digital elevation model (U.S. Geological Survey (USGS) NationalElevation Dataset). Catchment slope (degrees), aspect, and elevation were estimated as means for the contri-buting area upstream of the sampling point. Aspect was transformed before analysis as�1*cos(aspect-45) tocapture a southwest-northeast gradient representingmaximum andminimum inputs of solar radiation in thenorthern hemisphere. Streams were extracted from the digital elevation model (DEM) using the algorithm ofPlanchon and Darboux [2002]. Specific contributing area was derived using the D8 flow algorithm[O’Callaghan and Mark, 1984] with a channelization threshold of 20,000m2. This threshold was selectedbased on visual assessment of a ground-truthed stream layer including portions of the Poker, Caribou, andChatanika basins.

The areal extent of probable topographic depressions within each catchment, hereafter defined as wetlands,was estimated using a Monte Carlo approach executed with the Whitebox toolset (v. 3.2.1) [Creed et al., 2008].In this approach, random errors were added to DEM pixels, with values drawn from a normal distribution withmean of zero and standard deviation equal to the vertical accuracy of the DEM (1.84m), and a range of2000m estimated from a semivariogram to constrain spatial autocorrelation. The depression-filling algorithmofWang and Liu [2006] was then applied to the modified DEMs, and filled pixels were flagged during each of100 iterations. The depression probability for each pixel was calculated as the frequency of filling over the setof iterations. Probability thresholds of 5 to 55% were assessed in 5% increments and tested against lake andpond features extracted from the USGS National Hydrography Dataset. Wetland area in each catchment wasestimated by summing the area of pixels filled at greater than 30% probability. As an additional metric of thepresence of wetlands, cover by wetland-associated plant species was estimated from a Landsat-derived, 30mresolution data set (National Land Cover Dataset, 2001). Estimates of wetland area and cover by wetlandvegetation were correlated (r=0.64; Table 2).



Permafrost, soil, and additional vegetative attri-butes of catchments were estimated using simula-tions from the Terrestrial EcosystemModel (TEM), aprocess-based model designed to simulate C andN dynamics and exchange among vegetation, soil,and the atmosphere [McGuire et al., 1992; Raichet al., 1991]. The Dynamic Organic Soil version ofTEM (DOS-TEM) used in the present analysis wasdeveloped to represent the processes drivingspatial heterogeneity of terrestrial biogeochemicaland environmental dynamics in high-latitudeecosystems at a 1 km resolution [Yi et al., 2010].The model has been parameterized for six borealforest communities (upland and lowland blackspruce, white spruce, and deciduous boreal forest)[Yuan et al., 2012] and for four tundra communities(shrub, graminoid, heath, and wet sedge tundra)[Euskirchen et al., 2014; Euskirchen et al., 2009].Rate-limitingparameters for gross primaryproduc-tivity, autotrophic and heterotrophic respiration,maximum N uptake, N in litter, and soil C and Nimmobilization were calibrated until model valuesmatched available field-based estimates. Theseadjusted rate-limiting parameters and field-basedestimates were used to initialize themodel simula-tions. The model was applied at a 1 km resolutionover the spatial extent of the Yukon River Basin,which excluded the three highest latitude streamsincluded in the latitudinal gradient. Recent site-level validation of soil and vegetation C estimates[Yuan et al., 2012; Genet et al., 2015] and soil ther-mal and hydrological dynamics [Yi et al., 2009]has demonstrated the capacity of DOS-TEM toreproduce spatial and temporal heterogeneity inthe modeled attributes of terrestrial ecosystemswithin the study area.

Application of the model requires input data setsof vegetation distribution derived from theNorth American Land Change MonitoringSystem (www.cec.org/naatlas/nalcms), soil tex-ture, topography (elevation, slope, and aspect),atmospheric CO2 concentration, and climate,including monthly air temperature, precipitation,vapor pressure, and incoming shortwave radia-tion [Genet et al., 2013]. TEM-derived estimatesconsidered in regression models to predict streamchemistry included soil moisture (volumetricwater content) and temperature, which wereresolved for organic and mineral horizons, as wellas active layer depth, thickness of the soil organiclayer, and vegetation cover. TEM simulationsconsider a maximum soil depth of 5m; locationslacking permafrost are therefore assigned anTa

ble

2.Sp

earm

anCorrelatio

nCoe

fficien

tsAmon

gCatchmen

tAttrib

utes

Con

side

redin

Can

dida

teMultip

leRe

gression

Mod

els

Wetland

Extent

Aspect

Str.:Catch.

Leng

thCatch.

Area

Wetland

Veg

Dec.

Forest

Black

Spruce

Tund

raALD

Org.Layer

Thick.

VWC

May

VWC

July

VWC

Aug

SoilTemp

May

SoilTemp

July

SoilTemp

Aug

Slop

e�0

.73

0.01

0.32

0.48

�0.87

�0.09

�0.27

0.43

0.37

�0.25

0.51

0.64

0.63

�0.65

�0.66

�0.69

Wetland

extent

0.04

0.19

0.00

0.64

0.18

0.11

�0.01

0.00

0.09

�0.46

�0.67

�0.63

0.45

0.61

0.66

Aspect

�0.28

�0.17

�0.18

�0.03

�0.06

�0.02

0.30

�0.03

�0.04

�0.16

�0.13

0.12

0.19

0.17

Str.:catch.

leng

th0.90

�0.22

0.39

�0.11

0.41

0.29

�0.13

�0.09

�0.10

�0.06

0.00

0.07

0.08

Catch.area

�0.33

0.32

�0.02

0.53

0.30

�0.13

0.12

0.10

0.14

�0.12

�0.08

�0.06

Wetland

veg

0.08

0.44

�0.44

�0.33

0.42

�0.57

�0.43

�0.44

0.65

0.50

0.54

Deciduo

usforest

�0.20

�0.37

�0.47

�0.08

�0.37

�0.34

�0.44

0.52

0.41

0.35

Blackspruce

�0.02

�0.36

0.87

0.01

0.17

0.18

0.10

�0.05

0.05

Tund

ra0.49

�0.35

0.74

0.43

0.61

�0.66

�0.38

�0.33

ALD

�0.43

0.09

�0.07

0.12

�0.18

0.00

0.01

Org.layer

thickn

ess

�0.31

�0.01

�0.04

0.26

0.04

0.11

VWCMay

0.76

0.80

�0.84

�0.68

�0.65

VWCJuly

0.94

�0.77

�0.89

�0.84

VWCAug

�0.76

�0.80

�0.73

SoilTempMay

0.88

0.87

SoilTempJuly

0.98



www.cec.org/naatlas/nalcms

active layer depth of 5m. Soil temperature and moisture were summarized as monthly means for the studymonths of May, July, and August, and active layer depth represents the maximum annual thaw depth.TEM-derived attributes were weighted by area within each catchment to produce weighted-averagevalues for each attribute within each catchment (catchment area range: 43–5700 km2). We summarizedthese attributes both for 10 years preceding the study year (2009), and for 2009 only; however, attributevalues were qualitatively similar between the two time periods, and therefore, 2009 model output wasused in statistical analyses.

2.5. Statistical Methods

All statistical analyses except for the previously described PARAFAC modeling were conducted in R v. 3.0.2.Spearman rank correlations were used to assess relationships among constituents measured in streams, aswell as among the characteristics of catchments considered as candidate predictors in subsequent regressionanalyses of stream chemistry.

Multiple regression was used to identify significant predictors of stream chemistry. Data were centered bysubtracting the mean and dividing by standard deviation prior to analysis. Raw observations are presentedin graphics and summarized in Table 1. An initial set of candidate models was produced using the “all sub-sets” routine in package leaps, with model size limited to four or fewer predictor variables, followed by visualinspection of bivariate relationships for linearity. All candidate models were then inspected for multicollinear-ity of predictor variables, andmodels with variance inflation factors>2, or correlation coefficients>|0.5| wereremoved from the set of candidates (Table 2). Residuals of remaining models were assessed for normality,homoscedasticity, and points with high leverage, and transformations were undertaken to address theseissues. Model selection was based on Akaike’s information criterion (AIC), corrected for small sample size.For each response variable, we present the best set of models encompassing a “95% confidence set” basedon AIC weights.

3. Results3.1. Seasonal and Spatial Patterns in Stream Solutes

Nitrate concentration was highest in the southernmost streams, declined sharply beginning at the Tatalina River,and remained low until increasing at approximately 67° latitude (Figure 2a). Concentration of NO3

� declinedagain in the streams draining tundra in the northern foothills of the Brooks Range. These spatial patterns wereremarkably consistent across the three sampling times; however, in general, streams north of 67° latitude showedincreased NO3

� concentration in July and August compared to May (Figure 2a). Concentration of DOC showedan inverse latitudinal pattern compared to NO3

� (Figure 2b and Tables 4 and 4). In contrast to NO3�, DOC

concentration was seasonal, with highest concentration in all streams during snowmelt (Figure 2b).

PARAFAC modeling of EEMs yielded four components (C1–C4; Table 3). The components shared excitationand emission characteristics with components previously reported for other waters. C1 accounted for amajority of the fluorescent DOM pool and corresponded with previously studied humic, terrestrially derivedorganic material (Table 3). C1 was negatively correlated during all sampling times with FI, which indexes agradient of plant to microbial origin of compounds contributing to the DOM pool (Tables 4 and 5). Further,in July C1 was negatively correlated with the slope ratio, an index of meanmolecular weight of the DOM pool(Table 5), and covaried with DOC concentration during July and August (Table 5). Thus, higher values of C1correspond to DOM of terrestrial origin and humic composition, and in July the absorbance-based slope ratiofurther indicates that high relative contribution of component C1 co-occurs with DOM of higher molecularweights. Components C2 and C3 resembled previously studied oxidized and reduced quinones, respectively(Table 3). C2 and C3 components covaried negatively and positively with DOC concentration, respectively,and are not considered further here. Excitation and emission attributes of C4 were similar to those of theamino acid tryptophan, also referred to in previous studies as protein-like components (Table 3). C4 was posi-tively correlated with FI in all months, and with the slope ratio in May and July (Tables 4 and 5). Therefore, theprotein-like fluorophores described by component C4 coincided with presence of compounds of lower mole-cular weight and lower abundance of humic compounds in the DOM pool.

Composition of DOM varied seasonally and spatially, with sharp contrasts in the degree of variation amongthe measured indices. C1, the terrestrial, humic-like component derived from PARAFAC analysis, contributed



most strongly to the DOM pool in the midlatitude catchments during all sampling campaigns, as well as tocatchments along the northern half of the latitude gradient during snowmelt (Figure 2d). Fluorescence indexvalues indicated a stronger contribution of terrestrially derived DOM during snowmelt at nearly all sites, withvalues of 1.5–1.6 (Figure 2c). In July and August, FI values of 1.6–1.8 indicated increasing contribution ofmicrobially derived compounds to the DOM pool, with highest values in August (Figure 2c). Spatial patternsin FI were not apparent, with the exception of highest contribution of autochthonous DOM observed in threestreams with low DOC concentration at approximately 67° latitude (Figure 2). The C4 (tryptophan-like) com-ponent derived from PARAFAC analysis constituted a greater fraction of the DOM pool in July and Augustcompared to May in all but one of the streams (Figure 2e). Slope ratio was lowest at all sites in May, indicatinghigher averagemolecular weight of DOM during snowmelt, and values were similar across all sites (Figure 2f).Strong spatial patterns in the slope ratio of DOM occurred in July and August, with highest values at the

Table 3. Key Fluorescence Characteristics (Excitation and Emission Maxima) of Components Identified by PARAFAC Analysis in This Study, With Comparison toSimilar Components Identified in Previous Studiesa

Component(This Study)

ExcitationMaximum

EmissionMaximum

Molecular Characteristicsand Source (Identifiedby Previous Studies)

Stedmon andMarkager[2005]

Cory andMcKnight[2005]

Yamashitaet al.[2008]

Fellmanet al.[2009]

Coble[2007]

1 320–340 (<260) 442 humic-like component,terrestrial origin

3 or 4 10 or 1 2 5 C

2 <250, 300 443 oxidized quinone; humic-likecomponent; forested streams

with wetlands[Stedmon et al., 2003]

1 Q2 1 1 A

3 390 (280) 460 reduced quinone of microbial origin; highmolecular weight; aromatic

SQ2

4 300 383 tryptophan-like 7 8 4 8 M

aNumbered or lettered components associated with previous studies retain notation assigned by the referenced studies.

Figure 2. (a–f) Latitudal and temporal patterns in solute concentration and DOM composition across the regional spatialextent. Locations of the Yukon River and Brooks Range are indicated in Figures 2c and 2f.



southern end of the gradient and in streams draining the southern foothills of the Brooks Range, indicatingthe presence of lower relative molecular weight compounds in these streams (Figure 2e). However, thisspatial pattern contrasted with spatial patterns of DOC concentration, such that lower average molecularweight of the DOM pool coincided with a smaller pool of DOC.

3.2. Influence of Catchment Characteristics on Stream Solutes

In all seasons, variation in concentration of NO3� across the latitude gradient was explained primarily by a posi-

tive correlation with catchment slope, and simple regression models including slope as the only predictorexplained >50% of the variation in concentration (Figure 3 and Table 6). Factors explaining DOC concentrationweremore diverse and variable across sampling dates than for NO3

�, with vegetation and climate-driven predic-tors explaining nearly equal variation in concentration compared to catchment slope inMay and August (Figure 4and Table 6). Upland tundra cover additionally occurred in the most likely models explaining DOC concentrationin May and August, but had a positive effect during May, and a negative effect in August. Active layer depth wasnegatively correlated with DOC concentration and was present in the most likely models on all sampling dates.Catchment slope was negatively correlated with DOC concentration during May and August, but in July, onlyclimate-associated variables of soil temperature and active layer depth entered into the most likely model.Models explained twice as much variation in DOC (R2adj> 0.80) across the latitude gradient in July and Augustthan in May (R2adj = 0.42).

Compared to solute concentrations, predictors of DOM composition were varied, more distinct acrossseasons, and in general, candidate models explained less variation than for solute concentrations (maximumR2adj = 0.60). Soil moisture was a negative correlate of FI in themost likely models in May and July (Figure 5 andTable 6), indicating greater contribution of plant-derived compounds to the DOM pool where soils werewetter. Lowland tundra cover was additionally negatively correlated with FI and soil temperature waspositively correlated with FI in the most likely models during May, patterns that reflect greater contributionof microbially derived compounds in warmer soils and in catchments with less tundra vegetation. In July,catchment slope was positively related to FI in the most likely model, explaining 20%more variance in FI thanmodels including only predictors associated with climate or vegetation. The C1 component of fluorescentDOM was negatively related to organic layer thickness during May, and negatively related to active layerdepth in July and August (Figure 6 and Table 6). Catchment slope additionally occurred within the 95% con-fidence set of models describing C1 during August and was negatively related to C1 (Figure 6 and Table 6).These patterns indicate greater contribution of humic molecules to the DOM pool, where the organic andactive layers were less developed. Variation in the C4 component of fluorescent DOM, representingtryptophan-like compounds, was explained in May by a negative relationship with soil moisture (Table 6).

Table 4. Spearman Correlation Coefficients for Stream Chemistry Attributes During Snowmelt (May), n = 22 for SoluteConcentrations, n = 14 for DOM Composition

NO3� DOC C1 C4 Slope Ratio FI SUVA

NO3� �0.58 �0.50 0.35 0.16 0.65 �0.36

DOC 0.37 �0.06 �0.12 �0.19 0.36C1 �0.46 �0.24 �0.70 0.68C4 0.53 0.55 �0.29Slope ratio �0.10 �0.08FI �0.49

Table 5. Spearman Correlation Coefficients for Stream Attributes in July and Augusta

NO3� DOC C1 C4 Slope Ratio FI SUVA

NO3� �0.70 �0.72 �0.20 0.623 0.16 0.33

DOC �0.47 0.88 0.11 �0.642 �0.24 �0.02C1 �0.45 0.85 �0.17 �0.730 �0.58 �0.01C4 �0.58 0.28 0.13 0.427 0.71 �0.31Slope ratio 0.36 �0.66 �0.77 0.00 0.42 �0.11FI �0.21 �0.14 �0.30 0.58 0.24 �0.18SUVA �0.30 0.53 0.73 �0.03 �0.66 �0.33

aJuly values shown in top right (italicized; n = 22 for solute concentrations, n = 19 for DOM composition) and August in lower left (n = 22).



In July, several equally likely models containing diverse predictors explained variation in C4, but organic layerthickness explained twice as much variation as models including soil temperature and vegetation cover(Table 6). Models included a negative relationship with thickness of the organic layer, negative relationshipwith cover of upland black spruce, and positive correlation with temperature of organic soils. No models ofC4 met the criteria for candidacy in August. The slope ratio, a proxy for molecular weight of DOM, was pre-dictable only in July and August (Table 6 and Figure 7). The most likely candidate models describing sloperatio of DOM included positive effects of catchment slope and active layer depth in both July and Augustand a positive effect of upland tundra cover in August, with models that included catchment slope explaininggreatest variation in the slope ratio. These patterns indicate that compounds of lower molecular weightcontribute to the DOM pool in streams draining steep catchments with development of a deeper active layerand presence of tundra vegetation. Finally, latitudinal variation in SUVA could not be explained by any of thecandidate predictor variables. Across variables describing the composition of DOM, characteristics of soils,and vegetation were more likely predictors than for concentration of bulk DOC.

4. Discussion

Contrasts in soil and stream chemistry across climate and permafrost gradients have been interpreted asspace-for-time proxies to predict potential responses of carbon and nitrogen cycles to climate warming,

Figure 3. Significant predictors of NO3� concentration in streams along the latitude gradient. See Table 6 for regression

models. Note that multiple regression models were fit to centered data.



and in particular to thawing permafrost [Frey et al., 2007; Jones et al., 2005; Kawahigashi et al., 2004; Roth et al.,2013]. However, significant variation in the magnitude and direction of these biogeochemical patterns acrosshigh-latitude regions currently limits mechanistic understanding of observed patterns and predictive powerof ecosystemmodels that estimate material fluxes from catchments to coasts. Here we provide evidence thatin addition to the climate-driven attributes of active layer depth and soil temperature, catchment character-istics including mean slope, vegetation composition, and thickness of the soil organic layer influence the bio-geochemistry of catchments underlain by permafrost in northern Alaska. These results indicate thatcatchment setting interacts with attributes of permafrost to influence the composition of bioreactive solutesexported from catchments, and underscore the notion that accurately predicting and interpreting responsesof elemental cycles to changing climate, including thawing permafrost, must account for variation in physicaland vegetative attributes across catchments [McClelland et al., 2015].

Table 6. Multiple Regression Models of Catchment Attributes on Stream Chemistrya

Response Season Predictors Model n Adj R2 AICc Δ AICc Akaike Weight (wi)

NO3� May slope y = 0.829 * slope� 0.459 20 0.547 49.9 0.00 0.843

NO3� May ALD y = 0.424 * ALD� 0.269 18 0.137 53.3 3.37 0.157

NO3�b July slope y = 0.970 * slope� 0.517 20 0.636 49.2 0.00 0.672

NO3�b July To y =�0.5901 * To� 0.292 18 0.301 50.6 1.43 0.328

NO3–c August slope y = 0.384 * slope + 0.578 21 0.599 15.6 0.00 0.981

DOC May slope + utundra y =�0.883 * slope + 0.128 * utundra 17 0.419 48.3 0.00 0.717DOC May slope + utundra + ALD y =�0.301 * ALD + 0.246 * utundra� 0.830 * slope + 0.170 17 0.419 50.5 2.24 0.234DOCb July To + ALD y =�0.827 * ALD + 0.766 * To� 0.561 18 0.821 33.1 0.00 0.963DOCc August utundra + slope y =�0.346 * slope� 0.189 * utundra + 0.648 18 0.806 3.3 0.00 0.497DOCc August ALD + VWCo y =�0.265 * ALD� 0.356 * VWCo + 0.583 18 0.806 3.3 0.01 0.495FI May VWCm y =�0.822 * VWCm+ 0.442 12 0.435 34.9 0.00 0.590FI May ltundra y =�0.581 * ltundra + 0.352 12 0.334 36.9 1.98 0.219FI May To y = 0.565 * To + 0.373 12 0.215 38.8 3.96 0.081FI May VWCm+ ltundra y =�0.174 * ltundra� 0.644 * VWCm+ 0.436 12 0.388 39.3 4.43 0.065FId July VWCm+ slope y = 0.299 * slope� 0.443 * VWCm – 0.349 14 0.556 25.9 0.00 0.536FId July VWCm y =�0.391 * VWCm� 0.169 15 0.367 28.5 2.55 0.150FId July ALD y = 0.372 * ALD� 0.145 15 0.327 29.4 3.47 0.095FId July OL + ALD y = 0.289 * ALD� 0.272 * OL� 0.164 15 0.406 30.1 4.21 0.065FId July VWCm+ ALD y = 0.204 * ALD – 0.257 * VWCm 15 0.389 30.5 4.63 0.053FId July OL y =�0.404 * OL� 0.180 15 0.242 31.1 5.24 0.039FId July uBS + VWCm y =�0.3302 * VWCminJ� 0.1019 * uBS� 0.169 15 0.300 32.0 6.07 0.026FI August OL y =�0.518 * OL� 0.097 18 0.202 53.9 0.00 0.952C1 May OL y =�0.564 * OL� 0.133 12 0.286 NAe NA NAC1 July ALD y =�0.708 * ALD – 0.075 14 0.566 33.2 0.00 0.974C1 August ALD y =�0.595 * ALD + 0.022 18 0.284 52.1 0.00 0.612C1 August slope y =�0.679 * slope 21 0.432 53.0 0.91 0.388C4 May VWCm y =�0.804 * VWCm+ 0.201 12 0.391 NA NA NAC4 July OL y =�0.679 * OL� 0.152 14 0.566 44.5 0.00 0.218C4 July To y = 0.720 * To + 0.195 14 0.203 44.7 0.24 0.193C4 July uBS y =�0.591 * uBS� 0.109 14 0.199 44.8 0.31 0.186C4 July To + uBS y =�0.477 * uBS + 0.584 * To + 0.102 14 0.320 45.4 0.85 0.143C4 July To +OL y =�0.528 * OL + 0.549 * To + 0.058 14 0.311 45.5 1.04 0.130C4 July VWCm y =�0.433 * VWCm� 0.078 14 0.127 46.0 1.52 0.102SR July ALD + slope y = 0.852 * ALD + 0.261 * slope - 0.048 15 0.589 42.5 0.00 0.872SR July ALD y = 0.713 * ALD + 0.125 16 0.429 46.4 3.88 0.125SR August slope + utundra y = 0.414 * slope +0.603 * utundra +0.022 17 0.605 44.2 0.00 0.600SR August utundra y = 0.770 * utundra + 0.099 17 0.510 45.7 1.56 0.275SR August ALD + utundra y = 0.238 * ALD + 0.652 * utundra + 0.099 17 0.521 47.6 3.46 0.106

aModels summarized represent 95% confidence sets estimated by Akaike weights over candidate models for each response and season. ALD = active layerdepth; To = temperature of shallow organic soil; utundra = upland tundra extent; VWCo = volumetric water content of shallow, organic soils; ltundra = lowland tun-dra extent; VWCm= volumetric water content of upper mineral horizon. The spatial extent of TEM did not include three catchments, reducing n for models includingTEM-derived predictors.

bln(y + 1) transformed.cln(y + 2) transformed.dOne outlier excluded.eNA, not applicable because the candidate set included only one model.



4.1. Catchment Influence on Stream Solutes

Catchment slope had a dominant influence on NO3� concentration in streams, explaining twice as much var-

iation as climate-associated predictors in May and July and emerging as the sole predictor of NO3� concen-

tration in August. Thus, the apparent latitudinal patterns in NO3� concentration can be explained by

dominant topographic features in the region, which are characterized by upland catchments at the southernend of the study area, lowland catchments adjacent to the Yukon River, increased slope in the southern foot-hills of the Brooks Range, and gentle slopes north of the Brooks Range. Slope influences the transit time ofwater and solutes through catchments, with rapid transit promoting more conservative export of NO3

�

and slower flows increasing the opportunity for uptake and reaction of NO3� [Clow and Sueker, 2000;

Creed et al., 2008; Watmough et al., 2004]. Importantly, catchments with low relief allow for development ofriparian zones and wetlands that promote anoxia and removal of NO3

� by denitrification. A positive correla-tion between NO3

� concentration and active layer depth supported the hypothesis that permafrost influ-ences biogeochemical cycles, but this correlation occurred in May, when thaw depths are shallow.However, soils begin to thaw during snowmelt, and this positive relationship might stem from variation inthe rate of thaw and strong stratification in soil N pools with depth, causing greater export of NO3

� via deepflowpaths in more deeply thawed catchments [Jones et al., 2005; Harms and Jones, 2012]. Thus, this analysisindicates that greatest future increases in NO3

� export in this region will occur from steep catchmentsundergoing thawing of permafrost.

Regional variation in DOC concentration was explained by similar factors as NO3�, indicating that DOC and

NO3� are regulated by similar hydrologic mechanisms, but concentration of DOC was additionally influenced

by composition of vegetation and climate-associated drivers. A negative relationship between catchmentslope and DOC concentration during snowmelt and autumn is consistent with previous studies linking transittime to export of organic material, with wetlands often constituting the most significant source of DOM to

Figure 4. Significant predictors of DOC concentration in streams along the latitude gradient. See Table 6 for regressionmodels. Note that multiple regressionmodelswere fit to centered data.



rivers [Ågren et al., 2008b; Gergel et al., 1999; Pellerin et al., 2004]. However, unlike NO3�, catchment slope did

not explain significant variation in DOC concentration in all seasons. Furthermore, in contrast to other studies[e.g., Creed and Beall, 2009], topographic indicators in addition to catchment slope were not as strongly cor-related with DOC concentration. Instead, a negative correlation of active layer depth with DOC concentrationduring all sampling campaigns reflects the strong contribution of shallow organic layers to DOC in runoff andsuggests that deeper flows where permafrost has thawed will result in decreased export of DOC to streams,as has been observed in the Yukon basin [Striegl et al., 2005] and the Kuparuk basin in arctic Alaska[McClelland et al., 2007]. Extent of upland tundra vegetation also explained regional-scale variation in DOCconcentration, although a positive effect during snowmelt compared to a negative effect in autumn likelystems from strong seasonality of catchment hydrology, in addition to a direct influence of vegetation onproduction of DOC. Upland tundra typically occurs at higher latitudes or higher elevations where snowmeltoccurs later, and sampling was more likely to overlap with peak runoff and DOC concentration in thesecatchments (Figure S1). Lower productivity of high-elevation and high-latitude catchments compared toforested catchments likely results in a smaller source of DOC during the snow-free period.

A strong influence of vegetation characteristics on several descriptors of the DOM pool indicates the contri-bution of vegetation sources to DOM exported by streams, and a high likelihood of change in the composi-tion of DOM under the predicted scenarios of changing vegetation communities in boreal and arcticcatchments [Euskirchen et al., 2009; Johnstone et al., 2010; Tape et al., 2006]. In particular, correlations ofDOM composition with cover of black spruce and tundra suggest that replacement of these communitiesby deciduous forest or shrub cover could yield changes in the relative abundance of protein-like DOM (C4)and the average molecular weights (slope ratio) of DOM in streams. Additionally, correlations of active layerdepth with FI, C1, and the slope ratio suggest that the routing of water through deeper soils, as mediatedby climate, results in reduced export of high molecular weight, humic, and plant-derived molecules to

Figure 5. Significant predictors of FI (fluorescence index) in streams along the latitude gradient. See Table 6 for regression models. Note that multiple regressionmodels were fit to centered data.



streams. A positive relationship between FI and volumetric water content and soil temperature and anegative relationship of soil moisture with C4 (highly labile, tryptophan-like component of DOM) mightreflect increased microbial processing of DOM in warmer or wetter soils. Similar patterns have beenobserved in other boreal catchments, where higher FI occurred in forested catchments relative towetland-dominated catchments [Kothawala et al., 2015]. Further, saturated wetland soils were alsonegatively correlated with contribution of the tryptophan-like component in temperate catchments[Williams et al., 2010], which may be preferentially consumed by decomposers [Balcarczyk et al., 2009;Fellman et al., 2009]. Additional variation in the composition of DOM in this region that was unexplainedby the present analyses might be explained by groundwater contribution to streamflow, with more labile,but lower concentration of DOM contributed to streams by groundwater [O’Donnell et al., 2012].

Despite potential for temperature to influence the production of solutes via release from thawing permafrostor increased rate of primary production, soil temperature explained less variation in concentrations of NO3

�

and DOC, or composition of DOM than vegetation or catchment attributes. Previous experimental andcorrelative approaches at soil core, plot, and catchment scales have shown that higher temperature can yieldincreased production and concentration of soluble, microbially available N and DOM due to increasedrates of net primary production, increased solubilization of DOM, or more rapid turnover of C and N[Frey and Smith, 2005; Hobbie, 1996; Judd and Kling, 2002; Neff and Hooper, 2002; Schelker et al., 2013;

Figure 6. Significant predictors of the C1 component of DOM determined from PARAFAC analysis of EEMs. See Table 6 forregression models. Note that multiple regression models were fit to centered data.



Sponseller et al., 2014]. However, several studies of boreal and tundra soils have indicated that soil tem-perature effects on production of DOM are secondary to vegetation composition [Judd and Kling, 2002;Neff and Hooper, 2002] or soil moisture conditions [Schelker et al., 2013] and that higher soil tempera-tures enhance respiration more than production of DOM, resulting in little net effect on export ofDOM [Neff and Hooper, 2002]. Thus, whereas temperature influences mobilization and constrains exportof dissolved C and N at the plot scale, at broader scales the attributes of catchments may supersede thedirect effect of soil temperature.

4.2. Seasonality of Nitrate and Dissolved Organic Matter Across Latitudes

Seasonal patterns in DOC concentration and DOM composition observed in the study streams were consis-tent with those observed in other high-latitude regions, suggesting a common set of mechanisms influen-cing production and transport of DOC. The strong pulse of DOC associated with snowmelt observed acrossthe latitude gradient is a characteristic event in snowmelt-dominated catchments [Boyer et al., 1997; Caiet al., 2008; Holmes et al., 2012; Petrone et al., 2006; Sebestyen et al., 2008; Townsend-Small et al., 2011].DOM produced during snowmelt in the study catchments tended to be more humic and of higher molecularweight, as evidenced by lower FI values, higher values of the C1 component of EEMs, and lower values of theslope ratio during snowmelt compared to summer and autumn, and these traits are consistent with a detritalsource of DOM transported via shallow flowpaths [Carey, 2003; Neff et al., 2006; Spencer et al., 2008]. Similartemporal contrasts in both concentration of DOC and composition of DOM occurred across nearly all catch-ments, despite occurrence of sampling prior to peak discharge in the northernmost catchments and on thereceding limb of the snowmelt hydrograph at the southern end of the latitude gradient (Figure S1). However,the timing of sampling during snowmelt did correspond to a stronger contribution of terrestrially derivedDOM in streams sampled before peak discharge occurred compared to those sampled on the falling limb,as evidenced by higher contribution of the C1 component of fluorescent DOM in streams sampled prior topeak flow. This pattern suggests potential limitation by the size of the reservoir of readily mobilized humicmaterial or that rapid deepening of flowpaths through catchments during snowmelt result in bypassing ofthis pool during later stages of melt, a mechanism that generates hysteresis in solute-discharge relationships

Figure 7. Significant predictors of the slope ratio of DOM (SR; log-transformed ratio of the slope of absorbances at wavelengths 275–295 nm and 350–400 nm) instreams along the latitude gradient. See Table 6 for regression models. Note that multiple regression models were fit to centered data.



in many snowmelt-dominated catchments [Ågren et al., 2008b; Carey, 2003; Creed and Band, 1998]. The ulti-mate fate of the high load of dissolved C generated by snowmelt is unknown, but incubation studies indicatehigh relative bioavailability of DOC during snowmelt [Ågren et al., 2008a; Holmes et al., 2008], and potential forphotolysis because solar radiation is high during this time [Cory et al., 2014]. Depending upon transit times,these processes may be of importance for C cycling within freshwater ecosystems or in receiving estuaries.

In summer, lower concentration of DOC and evidence of lower relative concentrations of large, humic,plant-derived compounds were consistent with depletion of these materials from the leachable pool inorganic soils, or to a shift in the depth of flow connecting soils to streams. Both fluorescence- (FI and theC1 component) and absorbance-based indices (slope ratio) indicated this seasonal shift in the compositionof DOM (Figure 2). A deeper water table, and less frequent hydrologic connection of shallow, organic soilhorizons with streams than during snowmelt yields similar temporal patterns in concentration and compo-sition of DOM in other boreal ecosystems and in the temperate zone [Laudon et al., 2011; Sebestyen et al.,2008]. In permafrost-influenced catchments, a decline in concentration of DOC in summer is also likely dri-ven by increasing thaw depth, which allows flow through mineral soils that can sorb DOM and contain lesssoluble organic carbon than surface soils [Carey, 2003; Kawahigashi et al., 2006; Petrone et al., 2006].Although most indices of DOM composition were similar between July and August, SUVA values declinedin August at most sites, indicating decreased concentration of phenolic compounds in streams, despitesimilar flow conditions during the two sampling periods (Figure S1). Decreased aromaticity might occurbecause of increased microbial biomass and decomposition rates during warmer months or lower inputsof soil-derived DOM and longer processing time due to deeper flowpaths with increasing thaw depth[Battin et al., 2008; Inamdar et al., 2011; Mann et al., 2012; O’Donnell et al., 2010].

In contrast to DOM, the observed high-latitude streams exhibited little seasonal pattern in NO3�. Despite

sharp seasonal contrasts in depth of predominant flowpaths indicated by changes in the composition ofthe DOM pool, concentration of NO3

� remained relatively constant in each stream, suggesting that transportof NO3

� to streams is uninfluenced by seasonal changes in flow path depth or that multiple sources of NO3�

contribute to exports in streamflow. In temperate catchments, a flush of NO3� during snowmelt derives from

NO3� accumulated in shallow soils over winter, or direct delivery of atmospheric NO3

� from the snowpack tostreams [Creed and Band, 1998; Sebestyen et al., 2008]. Little overwinter production of NO3

� occurs in shallowarctic soils, but accumulation of ammonium suggests potential for nitrification in transit to streams[Buckeridge and Grogan, 2010; Schimel et al., 2004]. Lack of a pulse of NO3

� during snowmelt in these catch-ments may reflect the timing of the sampling with respect to a brief pulse of snow-derived NO3

�.Alternatively, highest rates of denitrification and NO3

� uptake occur in shallow soils of the study region dur-ing snowmelt [Harms and Jones, 2012], and thus, snow-derived NO3

� may be retained or removed beforereaching the stream. Additionally, upwelling of deep groundwater could contribute NO3

� directly to streams,providing a temporally constant source of N to streams [O’Donnell and Jones, 2006; Petrone et al., 2006].Potential sources of NO3

� to streams in summer also include contribution of NO3� derived from nitrification

within catchments, when rate of nitrification is highest [Schimel et al., 2004].

4.3. Consequences of Catchment Variation for Biogeochemical Responses to Climate Warming

Previous studies have documented decreased DOC flux from high-latitude catchments undergoing perma-frost thaw and hypothesized that deepening of flowpaths that bypass organic layers increase opportunityfor decomposition [McClelland et al., 2007; Striegl et al., 2005], whereas others have predicted an increasein DOC release from catchments due to permafrost thaw [Frey and Smith, 2005; Laudon et al., 2012].Concurrently, climate warming is expected to produce increased flux of inorganic N due to rerouting ofhydrologic flowpaths [Frey and McClelland, 2009]. Our analyses of diverse catchments spanning discontinu-ous to continuous extent of permafrost from boreal to arctic Alaska confirm the importance of permafrost,here indexed by depth of the active layer, for export of C and inorganic N from terrestrial ecosystems tostreams. However, along the Alaskan latitudinal transect, spatial patterns in NO3

� and DOC concentrationsas well as descriptors of the content of the DOM pool were not monotonic, suggesting that broad regionaltrends in climate and permafrost are not the primary influence on catchment biogeochemistry within thesampled region. Rather, characteristics of individual catchments, including physical and vegetative attributes,strongly shaped the amount and composition of biologically labile materials transported by streams. Thepatterns reported herein therefore indicate that a catchment context, rather than a top-down climate view,



is needed to predict changes in export of dissolved C and N at regional scales. We emphasize catchmentslope, development of the organic layer, and composition of vegetation as attributes of catchments thatmay correlate with future changes in exports of DOC and NO3

�, and accordingly, we predict that DOC andNO3

� export from steeply sloped catchments and those undergoing changes in vegetation compositionare likely to respond most strongly to thawing permafrost. Lack of correlation of DOC and NO3

� concentra-tion or DOM composition with catchment area suggests that these attributes are applicable across a widerange in catchment size. Importantly, a catchment context will support efforts to understand and model Cand N cycles in regions undergoing loss of permafrost.

ReferencesÅgren, A., M. Berggren, H. Laudon, and M. Jansson (2008a), Terrestrial export of highly bioavailable carbon from small boreal catchments in

spring floods, Freshwater Biol., 53(5), 964–972.Ågren, A., I. Buffam, M. Berggren, K. Bishop, M. Jansson, and H. Laudon (2008b), Dissolved organic carbon characteristics in boreal streams in

a forest-wetland gradient during the transition between winter and summer, J. Geophys. Res., 113, G03031, doi:10.1029/2007JG000674.Andersson, C. A., and R. Bro (2000), The N-way Toolbox for MATLAB, Chemom. Intell. Lab. Syst., 52(1), 1–4.Balcarczyk, K. L., J. B. Jones, R. Jaffé, and N. Maie (2009), Stream dissolved organic matter bioavailability and composition in watersheds

underlain with discontinuous permafrost, Biogeochemistry, 94(3), 255–270.Battin, T., L. Kaplan, S. Findlay, C. Hopkinson, E. Marti, A. Packman, J. Newbold, and F. Sabater (2008), Biophysical controls on organic carbon

fluxes in fluvial networks, Nat. Geosci., 1(2), 95–100.Boyer, E. W., G. M. Hornberger, K. E. Bencala, and D. M. McKnight (1997), Response characteristics of DOC flushing in an alpine catchment,

Hydrol. Processes, 11, 1635–1647.Buckeridge, K. M., and P. Grogan (2010), Deepened snow increases late thaw biogeochemical pulses in mesic low arctic tundra,

Biogeochemistry, 101(1–3), 105–121.Cai, Y. H., L. D. Guo, and T. A. Douglas (2008), Temporal variations in organic carbon species and fluxes from the Chena River, Alaska,

Limnol. Oceanogr., 53(4), 1408–1419.Carey, S. K. (2003), Dissolved organic carbon fluxes in a discontinuous permafrost subarctic alpine catchment, Permafrost Periglacial Process.,

14(2), 161–171.Carey, S. K., and W. L. Quinton (2005), Evaluating runoff generation during summer using hydrometric, stable isotope and hydrochemical

methods in a discontinuous permafrost alpine catchment, Hydrol. Processes, 19(1), 95–114.Carmack, E. C., R. W. Macdonald, and S. Jasper (2004), Phytoplankton productivity on the Canadian Shelf of the Beaufort Sea,Mar. Ecol. Prog. Ser.,

277, 37–50.Chapin, F. S., G. R. Shaver, A. E. Giblin, K. J. Nadelhoffer, and J. A. Laundre (1995), Responses of arctic tundra to experimental and observed

changes in climate, Ecology, 76, 694–711.Clow, D. W., and J. K. Sueker (2000), Relations between basin characteristics and stream water chemistry in alpine/subalpine basins in Rocky

Mountain National Park, Colorado, Water Resour. Res., 36(1), 49–61.Coble, P. G. (2007), Marine optical biogeochemistry: The chemistry of ocean color, Chem. Rev., 107(2), 402–418.Cory, R. M., and D. M. McKnight (2005), Fluorescence spectroscopy reveals ubiquitous presence of oxidized and reduced quinones in dissolved

organic matter, Environ. Sci. Technol., 39, 8142–8149.Cory, R. M., M. P. Miller, D. M. McKnight, J. Guerard, and P. L. Miller (2010), Effect of instrument-specific response on the analysis of fulvic acid

fluorescence spectra, Limnol. Oceanogr. Methods, 8, 67–78.Cory, R. M., C. P. Ward, B. C. Crump, and G. W. Kling (2014), Sunlight controls water column processing of carbon in arctic fresh waters, Science,

345(6199), 925–928.Creed, I. F., and F. D. Beall (2009), Distributed topographic indicators for predicting nitrogen export from headwater catchments,Water Resour.

Res., 45, W10407, doi:10.1029/2008WR007285.Creed, I. F., F. D. Beall, T. A. Clair, P. J. Dillon, and R. H. Hesslein (2008), Predicting export of dissolved organic carbon from forested catchments

in glaciated landscapes with shallow soils, Global Biogeochem. Cycles, 22, GB4024, doi:10.1029/2008GB003294.Creed, I., and L. Band (1998), Export of nitrogen from catchments within a temperate forest: Evidence for a unifying mechanism regulated by

variable source area dynamics, Water Resour. Res., 34, 3105–3120.Dillon, P. J., and L. A. Molot (1997), Effect of landscape form on export of dissolved organic carbon, iron, and phosphorus from forested

stream catchments, Water Resour. Res., 33(11), 2591–2600.Dornblaser, M., and R. Striegl (2007), Nutrient (N, P) loads at multiple scales and subbasin types in the Yukon River basin, Alaska, J. Geophys. Res.,

112, G04S57, doi:10.1029/2006JG000366.Euskirchen, E. S., A. D. McGuire, F. S. Chapin III, S. Yi, and C. C. Thompson (2009), Changes in vegetation in northern Alaska under scenarios of

climate change, 2003–2100: Implications for climate feedbacks, Ecol. Appl., 19(4), 1022–1043.Euskirchen, E. S., T. B. Carman, and A. D. McGuire (2014), Changes in the structure and function of northern Alaskan ecosystems when

considering variable leaf-out times across groupings of species in a dynamic vegetation model, Global Change Biol., 20(3), 963–978.Fellman, J. B., E. Hood, R. T. Edwards, and D. V. D’Amore (2009), Changes in the concentration, biodegradability, and fluorescent

properties of dissolved organic matter during stormflows in coastal temperate watersheds, J. Geophys. Res., 114, G01021,doi:10.1029/2008JG000790.

Frey, K. E., and J. W. McClelland (2009), Impacts of permafrost degradation on arctic river biogeochemistry, Hydrol. Processes, 23(1), 169–182.Frey, K. E., and L. C. Smith (2005), Amplified carbon release from vast West Siberian peatlands by 2100, Geophys. Res. Lett., 32, L09401,

doi:10.1029/2004GL022025.Frey, K., J. McClelland, R. Holmes, and L. Smith (2007), Impacts of climate warming and permafrost thaw on the riverine transport of nitrogen

and phosphorus to the Kara Sea, J. Geophys. Res., 112, G04S58, doi:10.1029/2006JG000369.Genet, H., et al. (2013), Modeling the effects of fire severity and climate warming on active layer thickness and soil carbon storage of black

spruce forests across the landscape in interior Alaska, Environ. Res. Lett., 8(4), 045016.Genet, H., Y. He, A. D. McGuire, Q. Zhuang, Y. Zhang, F. Biles, D. D’Amore, X. Zhou, and K. Johnson (2015), Terrestrial carbon modeling:

Baseline and projections in upland ecosystems Rep., USGS.



AcknowledgmentsData are available at http://www.lter.uaf.edu/data_detail.cfm?datafile_p-key=456. We thank Ben Abbott, AmandaRinehart, Kurt Smart, and Jessie Young fortheir assistance with sample collection.We acknowledge H. Laudon and twoanonymous reviewers for comments thatimproved the manuscript. This work wassupported by the National ScienceFoundation (OPP-0817056, to T.K.H.).

http://dx.doi.org/10.1029/2007JG000674

http://dx.doi.org/10.1029/2008WR007285

http://dx.doi.org/10.1029/2008GB003294

http://dx.doi.org/10.1029/2006JG000366

http://dx.doi.org/10.1029/2008JG000790

http://dx.doi.org/10.1029/2004GL022025

http://dx.doi.org/10.1029/2006JG000369

http://www.lter.uaf.edu/data_detail.cfm?datafile_pkey=456



Gergel, S. E., M. G. Turner, and T. K. Kratz (1999), Dissolved organic carbon as an indicator of the scale of watershed influence on lakes andrivers, Ecol. Appl., 9(4), 1377–1390.

Harden, J. W., et al. (2012), Field information links permafrost carbon to physical vulnerabilities of thawing, Geophys. Res. Lett., 39, L15704,doi:10.1029/2012GL051958.

Harms, T. K., and J. B. J. Jones (2012), Thaw depth determines reaction and transport of inorganic nitrogen in valley bottom permafrost soils,Global Change Biol., 18, 2958–2969.

Helms, J. R., A. Stubbins, J. D. Ritchie, E. C. Minor, D. J. Kieber, and K. Mopper (2008), Absorption spectral slopes and slope ratios as indicatorsof molecular weight, source, and photobleaching of chromophoric dissolved organic matter, Limnol. Oceanogr., 53(3), 955–969.

Hobbie, S. (1996), Temperature and plant species control over litter decomposition in Alaskan tundra, Ecol. Monogr., 66(4), 503–522.Holmes, R. M., J. W. McClelland, P. A. Raymond, B. B. Frazer, B. J. Peterson, andM. Stieglitz (2008), Lability of DOC transported by Alaskan rivers

to the Arctic Ocean, Geophys. Res. Lett., 35, L03402, doi:10.1029/2007GL032837.Holmes, R. M., et al. (2012), Seasonal and annual fluxes of nutrients and organic matter from large rivers to the Arctic Ocean and surrounding

seas, Estuaries Coasts, 35(2), 369–382.Inamdar, S., S. Singh, S. Dutta, D. Levia, M. Mitchell, D. Scott, H. Bais, and P. McHale (2011), Fluorescence characteristics and sources of

dissolved organic matter for stream water during storm events in a forested mid-Atlantic watershed, J. Geophys. Res., 116, G03043,doi:10.1029/2011JG001735.

Jaffe, R., D. McKnight, N. Maie, R. Cory, W. H. McDowell, and J. L. Campbell (2008), Spatial and temporal variations in DOM composition inecosystems: The importance of long-term monitoring of optical properties, J. Geophys. Res., 113, G04032, doi:10.1029/2008JG000683.

Johnstone, J. F., T. N. Hollingsworth, F. S. Chapin III, and M. C. Mack (2010), Changes in fire regime break the legacy lock on successionaltrajectories in Alaskan boreal forest, Global Change Biol., 16(4), 1281–1295.

Jones, J. B., K. C. Petrone, J. C. Finlay, L. D. Hinzman, and W. R. Bolton (2005), Nitrogen loss from watersheds of interior Alaska underlain withdiscontinuous permafrost, Geophys. Res. Lett., 32, L02401, doi:10.1029/2004GL021734.

Judd, K. E., and G. W. Kling (2002), Production and export of dissolved C in arctic tundra mesocosms: The roles of vegetation and water flow,Biogeochemistry, 60, 213–234.

Kawahigashi, M., K. Kaiser, K. Kalbitz, A. Rodionov, and G. Guggenberger (2004), Dissolved organic matter in small streams along a gradientfrom discontinuous to continuous permafrost, Global Change Biol., 10(9), 1576–1586.

Kawahigashi, M., K. Kaiser, A. Rodionov, and G. Guggenberger (2006), Sorption of dissolved organic matter by mineral soils of the Siberianforest tundra, Global Change Biol., 12(10), 1868–1877.

Keller, K., J. D. Blum, and G. W. Kling (2010), Stream geochemistry as an indicator of increasing permafrost thaw depth in an arctic watershed,Chem. Geol., 273(1–2), 76–81.

Keuper, F., P. M. van Bodegom, E. Dorrepaal, J. T. Weedon, J. van Hal, R. S. P. van Logtestijn, and R. Aerts (2012), A frozen feast: Thawingpermafrost increases plant-available nitrogen in subarctic peatlands, Global Change Biol., 18, 1998–2007.

Kicklighter, D. W., D. J. Hayes, J. W. McClelland, B. J. Peterson, A. D. McGuire, and J. M. Melillo (2013), Insights and issues with simulatingterrestrial DOC loading of Arctic river networks, Ecol. Appl., 23(8), 1817–1836.

Kling, G. W., G. W. Kipphut, andM. C. Miller (1991), Arctic lakes and streams as gas conduits to the atmosphere: Implications for tundra carbonbudgets, Science, 251(4991), 298–301.

Kortelainen, P., T. Mattsson, L. Finer, M. Ahtiainen, S. Saukkonen, and T. Sallantaus (2006), Controls on the export of C, N, P and Fe fromundisturbed boreal catchments, Finland, Aquat. Sci., 68(4), 453–468.

Kothawala, D. N., X. Ji, H. Laudon, A. M. Âgren, M. N. Futter, S. J. Köhler, and L. J. Tranvik (2015), The relative influence of land cover, hydrology,and in-stream processing on the composition of dissolved organic matter in boreal streams, J. Geophys. Res. Biogeosci., 120, 1491–1505,doi:10.1002/2015JG002946.

Lafrenière, M. J., and S. F. Lamoureux (2013), Thermal perturbation and rainfall runoff have greater impact on seasonal solute loads thanphysical disturbance of the active layer, Permafrost Periglacial Process., 24(3), 241–251.

Laudon, H., M. Berggren, A. Ågren, I. Buffam, K. Bishop, T. Grabs, M. Jansson, and S. Kohler (2011), Patterns and dynamics of dissolved organiccarbon (DOC) in boreal streams: The role of processes, connectivity, and scaling, Ecosystems, 14, 880–893.

Laudon, H., J. Buttle, S. K. Carey, J. McDonnell, K. McGuire, J. Seibert, J. Shanley, C. Soulsby, and D. Tetzlaff (2012), Cross-regional prediction oflong-term trajectory of stream water DOC response to climate change, Geophys. Res. Lett., 39, L18404, doi:10.1029/2012GL053033.

Le Fouest, V., M. Babin, and J. E. Tremblay (2013), The fate of riverine nutrients on Arctic shelves, Biogeosciences, 10(6), 3661–3677.Lovett, G. M., K. C. Weathers, and M. A. Arthur (2002), Control of nitrogen loss from forested watersheds by soil carbon:nitrogen ratio and tree

species composition, Ecosystems, 5, 712–718.Mann, P. J., A. Davydova, N. Zimov, R. G. M. Spencer, S. Davydov, E. Bulygina, S. Zimov, and R. M. Holmes (2012), Controls on the composition

and lability of dissolved organic matter in Siberia’s Kolyma River basin, J. Geophys. Res., 117, G01028, doi:10.1029/2011JG001798.McClelland, J. W., A. Townsend-Small, R. M. Holmes, F. Pan, M. Stieglitz, M. Khosh, and B. J. Peterson (2014), River export of nutrients and

organic matter from the North Slope of Alaska to the Beaufort Sea, Water Resour. Res., 50, 1823–1839, doi:10.1002/2013WR014722.McClelland, J. W., S. E. Tank, R. G. M. Spencer, and A. I. Shiklomanov (2015), Coordination and sustainability of river observing activities in the

Arctic, Arctic, 68(5), 1–10.McClelland, J., M. Stieglitz, F. Pan, R. Holmes, and B. Peterson (2007), Recent changes in nitrate and dissolved organic carbon export from the

upper Kuparuk River, North Slope, Alaska, J. Geophys. Res., 112, G04S60, doi:10.1029/2006JG000371.McGuire, A. D., J. M. Melillo, L. A. Joyce, D. W. Kicklighter, A. L. Grace, B. I. Moore, and C. J. Vorosmarty (1992), Interactions between carbon and

nitrogen dynamics in estimating net primary productivity for potential vegetation in North America, Global Biogeochem. Cycles, 6, 101–124.McGuire, A. D., L. G. Anderson, T. R. Christensen, S. Dallimore, L. D. Guo, D. J. Hayes, M. Heimann, T. D. Lorenson, R. W. Macdonald, and

N. Roulet (2009), Sensitivity of the carbon cycle in the Arctic to climate change, Ecol. Monogr., 79(4), 523–555.McGuire, K. J., J. J. McDonnell, M. Weiler, C. Kendall, B. L. McGlynn, J. M. Welker, and J. Seibert (2005), The role of topography on

catchment-scale water residence time, Water Resour. Res., 41, W05002, doi:10.1029/2004WR003657.McKnight, D. M., E. W. Boyer, P. K. Westerhoff, P. T. Doran, T. Kulbe, and D. T. Andersen (2001), Spectrofluorometric characterization of

dissolved organic matter for indication of precursor organic material and aromaticity, Limnol. Oceanogr., 46, 38–48.Neff, J. C., and D. U. Hooper (2002), Vegetation and climate controls on potential CO2, DOC and DON production in northern latitude soils,

Global Change Biol., 8(9), 872–884.Neff, J. C., J. C. Finlay, S. A. Zimov, S. P. Davydov, J. J. Carrasco, E. A. G. Schuur, and A. I. Davydova (2006), Seasonal changes in the age and

structure of dissolved organic carbon in Siberian rivers and streams, Geophys. Res. Lett., 33, L23401, doi:10.1029/2006GL028222.O’Callaghan, J. F., andD. M. Mark (1984), The extraction of drainage networks fromdigital elevation data, Comput. Vision, Graphics, Image Process.,

28, 323–344.



http://dx.doi.org/10.1029/2012GL051958

http://dx.doi.org/10.1029/2007GL032837

http://dx.doi.org/10.1029/2011JG001735

http://dx.doi.org/10.1029/2008JG000683

http://dx.doi.org/10.1029/2004GL021734

http://dx.doi.org/10.1002/2015JG002946

http://dx.doi.org/10.1029/2012GL053033

http://dx.doi.org/10.1029/2011JG001798

http://dx.doi.org/10.1002/2013WR014722

http://dx.doi.org/10.1029/2006JG000371

http://dx.doi.org/10.1029/2004WR003657

http://dx.doi.org/10.1029/2006GL028222

O’Donnell, J. A., and J. B. Jones (2006), Nitrogen retention in the riparian zone of catchments underlain by discontinuous permafrost,Freshwater Biol., 51(5), 854–864.

O’Donnell, J. A., G. R. Aiken, E. S. Kane, and J. B. Jones (2010), Source water controls on the character and origin of dissolved organic matter instreams of the Yukon River basin, Alaska, J. Geophys. Res., 115, G03025, doi:10.1029/2009JG001153.

O’Donnell, J. A., G. R. Aiken, M. A. Walvoord, and K. D. Butler (2012), Dissolved organic matter composition of winter flow in the Yukon Riverbasin: Implications of permafrost thaw and increased groundwater discharge, Global Biogeochem. Cycles, 26, GB0E06, doi:10.1029/2012GB004341.

Pellerin, B. A., W. M. Wollheim, C. S. Hopkinson, M. H. McDowell, M. R. Williams, C. J. Vorosmarty, and M. L. Daley (2004), Role of wetlands anddeveloped land use on dissolved organic nitrogen concentrations and DON/TDN in northeastern US rivers and streams,Limnol. Oceanogr., 49, 910–918.

Petrone, K. C., J. B. Jones, L. D. Hinzman, and R. D. Boone (2006), Seasonal export of carbon, nitrogen, and major solutes from Alaskancatchments with discontinuous permafrost, J. Geophys. Res., 111, G02020, doi:10.1029/2005JG000055.

Planchon, O., and F. Darboux (2002), A fast, simple and versatile algorithm to fill the depressions of digital elevationmodels, Catena, 46, 159–176.Raich, J. W., E. B. Rastetter, J. M. Melillo, D. W. Kicklighter, P. A. Steudler, B. J. Peterson, A. L. Grace, B. Moore, and C. J. Vorosmarty (1991),

Potential net primary productivity in South America - Application of a global model, Ecol. Appl., 1(4), 399–429.Roth, V.-N., T. Dittmar, R. Gaupp, and G. Gleixner (2013), Latitude and pH driven trends in the molecular composition of DOM across a north

south transect along the Yenisei River, Geochim. Cosmochim. Acta, 123, 93–105.Schelker, J., T. Grabs, K. Bishop, and H. Laudon (2013), Drivers of increased organic carbon concentrations in stream water following forest

disturbance: Separating effects of changes in flow pathways and soil warming, J. Geophys. Res. Biogeosci., 118, 1814–1827, doi:10.1002/2013JG002309.

Schimel, J. P., C. Bilbrough, and J. A. Welker (2004), Increased snow depth affects microbial activity and nitrogen mineralization in two Arctictundra communities, Soil Biol. Biochem., 36(2), 217–227.

Sebestyen, S. D., E. W. Boyer, J. B. Shanley, C. Kendall, D. H. Doctor, G. R. Aiken, and N. Ohte (2008), Sources, transformations, and hydrologicalprocesses that control stream nitrate and dissolved organic matter concentrations during snowmelt in an upland forest,Water Resour. Res., 44, W12410, doi:10.1029/2008WR006983.

Spencer, R. G. M., G. R. Aiken, K. P. Wickland, R. G. Striegl, and P. J. Hernes (2008), Seasonal and spatial variability in dissolved organic matterquantity and composition from the Yukon River basin, Alaska, Global Biogeochem. Cycles, 22, GB4002, doi:10.1029/2008GB003231.

Sponseller, R. A., J. Temnerud, K. Bishop, and H. Laudon (2014), Patterns and drivers of riverine nitrogen (N) across alpine, subarctic, andboreal Sweden, Biogeochemistry, 120(1–3), 105–120.

Stedmon, C. A., and R. Bro (2008), Characterizing dissolved organic matter fluorescence with parallel factor analysis: A tutorial, Limnol.Oceanogr. Methods, 6, 572–579.

Stedmon, C. A., and S. Markager (2005), Tracing the production and degradation of autochthonous fractions of dissolved organic matter byfluorescence analysis, Limnol. Oceanogr., 50, 1415–1426.

Stedmon, C. A., S. Markager, and R. Bro (2003), Tracing dissolved organic matter in aquatic environments using a new approach to fluor-escence spectroscopy, Mar. Chem., 82, 239–254.

Striegl, R. G., G. R. Aiken, M. M. Dornblaser, P. A. Raymond, and K. P. Wickland (2005), A decrease in discharge-normalized DOC export by theYukon River during summer through autumn, Geophys. Res. Lett., 32, L21413, doi:10.1029/2005GL024413.

Striegl, R. G., M. M. Dornblaser, C. P. McDonald, J. R. Rover, and E. G. Stets (2012), Carbon dioxide and methane emissions from the YukonRiver system, Global Biogeochem. Cycles, 26, GB0E05, doi:10.1029/2012GB004306.

Strohmeier, S., K. H. Knorr, M. Reichert, S. Frei, J. H. Fleckenstein, S. Peiffer, and E. Matzner (2013), Concentrations and fluxes of dissolvedorganic carbon in runoff from a forested catchment: Insights from high frequency measurements, Biogeosciences, 10(2), 905–916.

Tank, S. E., M. Manizza, R. M. Holmes, J. W. McClelland, and B. J. Peterson (2012), The processing and impact of dissolved riverine nitrogen inthe Arctic Ocean, Estuaries Coasts, 35(2), 401–415.

Tape, K., M. Sturm, and C. Racine (2006), The evidence for shrub expansion in Northern Alaska and the Pan-Arctic, Global Change Biol., 12(4),686–702.

Tarnocai, C., J. G. Canadell, E. A. G. Schuur, P. Kuhry, G. Mazhitova, and S. Zimov (2009), Soil organic carbon pools in the northern circumpolarpermafrost region, Global Biogeochem. Cycles, 23, GB2023, doi:10.1029/2008GB003327.

Townsend-Small, A., J. W. McClelland, R. M. Holmes, and B. J. Peterson (2011), Seasonal and hydrologic drivers of dissolved organic matterand nutrients in the upper Kuparuk River, Alaskan Arctic, Biogeochemistry, 103(1–3), 109–124.

Walker, S. A., R. M. W. Amon, and C. A. Stedmon (2013), Variations in high-latitude riverine fluorescent dissolved organic matter: A comparison oflarge Arctic rivers, J. Geophys. Res. Biogeosci., 118, 1689–1702, doi:10.1002/2013JG002320.

Walvoord, M. A., and R. G. Striegl (2007), Increased groundwater to stream discharge from permafrost thawing in the Yukon River basin:Potential impacts on lateral export of carbon and nitrogen, Geophys. Res. Lett., 34, L12402, doi:10.1029/2007GL030216.

Wang, L., and H. Liu (2006), An efficient method for identifying and filling surface depressions in digital elevation models for hydrologicanalysis and modeling, Int. J. Geogr. Inf. Sci., 20(2), 193–213.

Watmough, S. A., M. C. Eimers, J. Aherne, and P. J. Dillon (2004), Climate effects on stream nitrate concentrations at 16 forested catchments insouth central Ontario, Environ. Sci. Technol., 38(8), 2383–2388.

Weishaar, J. L., G. R. Aiken, B. A. Bergamaschi, M. S. Fram, R. Fujii, and K. Mopper (2003), Evaluation of specific ultraviolet absorbance as anindicator of the chemical composition and reactivity of dissolved organic carbon, Environ. Sci. Technol., 37(20), 4702–4708.

Williams, C. J., Y. Yamashita, H. F. Wilson, R. Jaffé, and M. A. Xenopoulos (2010), Unraveling the role of land use and microbial activity inshaping dissolved organic matter characteristics in stream ecosystems, Limnol. Oceanogr., 55(3), 1159–1171.

Yamashita, Y., R. Jaffé, N. Maie, and E. Tanoue (2008), Assessing the dynamics of dissolved organic matter (DOM) in coastal environments byexcitation emission matrix fluorescence and parallel factor analysis (EEM-PARAFAC), Limnol. Oceanogr., 53, 1900–1908.

Yarie, J., and K. Van Cleve (2010), Long-term monitoring of climatic and nutritional effects on tree growth in interior Alaska, Can. J. For. Res.,40(7), 1325–1335.

Yi, S., K. Manies, J. Harden, and A. D. McGuire (2009), Characteristics of organic soil in black spruce forests: Implications for the application ofland surface and ecosystem models in cold regions, Geophys. Res. Lett., 36, L05501, doi:10.1029/2008GL037014.

Yi, S., A. D. McGuire, E. Kasischke, J. Harden, K. Manies, M. Mack, and M. Turetsky (2010), A dynamic organic soil biogeochemical model forsimulating the effects of wildfire on soil environmental conditions and carbon dynamics of black spruce forests, J. Geophys. Res., 115,G04015, doi:10.1029/2010JG001302.

Yuan, F. M., S. H. Yi, A. D. McGuire, K. D. Johnson, J. Liang, J. W. Harden, E. S. Kasischke, and W. A. Kurz (2012), Assessment of boreal foresthistorical C dynamics in the Yukon River Basin: Relative roles of warming and fire regime change, Ecol. Appl., 22, 2091–2109.



http://dx.doi.org/10.1029/2009JG001153

http://dx.doi.org/10.1029/2012GB004341

http://dx.doi.org/10.1029/2012GB004341

http://dx.doi.org/10.1029/2005JG000055

http://dx.doi.org/10.1002/2013JG002309

http://dx.doi.org/10.1002/2013JG002309

http://dx.doi.org/10.1029/2008WR006983

http://dx.doi.org/10.1029/2008GB003231

http://dx.doi.org/10.1029/2005GL024413

http://dx.doi.org/10.1029/2012GB004306

http://dx.doi.org/10.1029/2008GB003327

http://dx.doi.org/10.1002/2013JG002320

http://dx.doi.org/10.1029/2007GL030216

http://dx.doi.org/10.1029/2008GL037014

http://dx.doi.org/10.1029/2010JG001302

Erratum

In the originally published version of this article, Table 6, column 4 contained several symbol errors. In thiscolumn, the letter b appeared in place of the correct symbol, an asterisk. These errors have since beencorrected, and this version may be considered the authoritative version of record.



catchment influence on nitrate and dissolved organic

Documents